9.1 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Objectives To describe the benefits of a virtual memory system To explain the concepts of demand paging, page- replacement algorithms, and allocation of page frame To discuss the principle of the working-set model Chapter 9: Virtual-Memory Management

9.2 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Chapter 9: Virtual-Memory Management 9.1 Background 9.2 Demand Paging 9.3 Copy-on-Write 9.4 Page Replacement 9.5 Allocation of Frames 9.6 Thrashing 9.7 Memory-Mapped Files (skip) 9.8 Allocating Kernel Memory 9.9 Other Considerations 9.10 Operating-System Examples (skip)

9.3 Silberschatz, Galvin and Gagne ©2005 Operating System Principles 9.1 Background Virtual memory – separation of user logical memory from physical memory. Only part of the program needs to be in memory for execution Logical address space can therefore be much larger than physical address space Allows address spaces to be shared by several processes Allows for more efficient process creation Virtual memory can be implemented via: Demand paging Demand segmentation

9.4 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Virtual Memory That is Larger Than Physical Memory  (page table)

9.5 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Virtual-address Space for functional calls for new objects

9.6 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Shared Library Using Virtual Memory

9.7 Silberschatz, Galvin and Gagne ©2005 Operating System Principles 9.2 Demand Paging Bring a page into memory only when it is needed Less I/O needed Less memory needed Faster response More users Page is needed  reference to it invalid reference  abort not-in-memory  bring to memory Lazy swapper – never swaps a page into memory unless that page will be needed A swapper manipulates entire processes A pager deals with pages of a process

9.8 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Transfer of a Paged Memory to Contiguous Disk Space

9.9 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Valid-Invalid Bit With each page table entry a valid–invalid bit is associated (v  in-memory, i  not-in-memory) Initially valid–invalid bit is set to i on all entries Example of a page table snapshot: During address translation, if valid–invalid bit in page table entry is i  page fault v v v v i i i …. Frame #valid-invalid bit page table

9.10 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Page Table When Some Pages Are Not in Main Memory

9.11 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Page Fault If there is a reference to a page, first reference to that page will trap to operating system: page-fault Steps in Handling a Page Fault: 1. Operating system looks at an internal table 2. If invalid reference  abort; If just not in memory  continue 3. Get an empty frame 4. Read the desired page into frame (swap in) 5. Reset the internal table (set validation bit = v ) and page table. 6. Restart the instruction that caused the page fault

9.12 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Steps in Handling a Page Fault

9.13 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Page Fault Pure demand paging: never bring a page into memory until it is required Multiple page faults per instruction possible? Fortunately, programs tend to have locality of reference A page fault in a three-address instruction (like z = x + y ) will require fetching the instruction again, decoding it again, fetching the two operands again, and then adding again.

9.14 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Page Fault Problem Restart instruction difficulty block move of 256 bytes (p. 365) Solutions  Attempts to access both block ends first  Uses temporary registers to hold values of overwritten locations. If page fault, then all the old values are written back into memory before the trap occurs. This restores memory into its state before the instruction.

9.15 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Performance of Demand Paging Page Fault Rate: 0  p  1.0 if p = 0, no page faults if p = 1, every reference is a fault Effective Access Time (EAT) EAT = (1 – p) x memory access + p ( page fault overhead + swap page out + swap page in + restart overhead )

9.16 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Demand Paging Example Memory access time = 200 nanoseconds Average page-fault service time = 8 milliseconds EAT = (1 – p) x p (8 milliseconds) = (1 – p ) x p x 8,000,000 = p x 7,999,800 If one access out of 1,000 causes a page fault, then EAT = 8.2 microseconds. This is a slowdown by a factor of 40!! If we want performance degradation less than 10% p x 7,999,800 < 220  p < (1 out of 399,990) To implement demand paging Need frame-allocation and page-replacement algorithm

9.17 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Demand Paging Handling and overall use of swap space Disk I/O to swap space is generally faster than that to the file system Better paging throughput can be gained  by copying an entire file image into the swap space at process startup and then perform demand paging from the swap space.  Or to demand pages from the file system initially but to write the pages to swap space as they are replaced. Some systems limit the amount of swap space used through demand paging of binary files. These frames can simply be overwritten.

9.18 Silberschatz, Galvin and Gagne ©2005 Operating System Principles 9.3 Copy-on-Write During process creation (check p. 113), Copy-on- Write (COW) allows both parent and child processes to initially share the same pages in memory If either process modifies a shared page, only then is the page copied COW allows more efficient process creation as only modified pages are copied Free pages are allocated from a pool of zeroed-out pages (all 0’s)

9.19 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Before Process 1 Modifies Page C After Process 1 Modifies Page C

9.20 Silberschatz, Galvin and Gagne ©2005 Operating System Principles 9.4 What happens if there is no free frame? If a process of ten pages actually uses only 5 pages, then demand paging saves the I/O to load the 5 pages never used. The degree of multi-programming could be increased twice. We may over-allocating memory. Buffers for I/O also consume a lot of memory. Page replacement – find some page in memory, but not really in use, swap it out want an page replacement algorithm which will result in minimum number of page faults Same page may be brought into memory several times

9.21 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Page Replacement Prevent over-allocation of memory by modifying page-fault service routine to include page replacement Use modify (dirty) bit to reduce overhead of page transfers – only modified pages are written to disk Page replacement completes separation between logical memory and physical memory – large virtual memory can be provided on a smaller physical memory

9.22 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Need For Page Replacement Section 9.6 consider reducing the level of multiprogramming

9.23 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Basic Page Replacement 1. Find the location of the desired page on disk 2. Find a free frame: - If there is a free frame, use it - If there is no free frame, use a page replacement algorithm to select a victim frame - Write the victim frame to the disk; change the page and frame tables 3. Bring the desired page into the (newly) free frame; update the page and frame tables 4. Restart the process

9.24 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Page Replacement

9.25 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Page Replacement Algorithms Want lowest page-fault rate Evaluate algorithm by running it on a particular string of memory references (reference string) and computing the number of page faults on that string In all our examples, the reference string is 1, 2, 3, 4, 1, 2, 5, 1, 2, 3, 4, 5

9.26 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Graph of Page Faults Versus The Number of Frames

9.27 Silberschatz, Galvin and Gagne ©2005 Operating System Principles First-In-First-Out (FIFO) Algorithm Reference string: 1, 2, 3, 4, 1, 2, 5, 1, 2, 3, 4, 5 3 frames (3 pages can be in memory at a time per process) 4 frames Belady’s Anomaly: more frames  more page faults page faults page faults 4 43

9.28 Silberschatz, Galvin and Gagne ©2005 Operating System Principles FIFO Page Replacement Exercise: Try the 4 frame case

9.29 Silberschatz, Galvin and Gagne ©2005 Operating System Principles FIFO Illustrating Belady’s Anomaly

9.30 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Optimal Algorithm Replace page that will not be used for longest period of time 4 frames example 1, 2, 3, 4, 1, 2, 5, 1, 2, 3, 4, 5 How do you know this? Used for measuring how well your algorithm performs page faults 4 Exercise: Try the 3 frame case

9.31 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Optimal Page Replacement

9.32 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Least Recently Used (LRU) Algorithm Reference string: 1, 2, 3, 4, 1, 2, 5, 1, 2, 3, 4, 5 Counter implementation Add to CPU a logical clock (counter); every page entry has a time-of-use field. every time page is referenced through this entry, copy the clock into the time-of-use field. When a page needs to be replaced, look at the time-of-use fields to determine which are to be replaced Exercise: Try the 3 frame case

9.33 Silberschatz, Galvin and Gagne ©2005 Operating System Principles LRU Page Replacement

9.34 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Stack implementation (to record the most recent page references) - keep a stack of page numbers in a double link form: Page referenced:  move it to the top  Use doubly-linked list: requires 6 pointers to be changed No search for replacement

9.35 Silberschatz, Galvin and Gagne ©2005 Operating System Principles LRU Approximation Algorithms Reference bit With each page associate a bit, initially = 0 When page is referenced bit set to 1 Replace the one which is 0 (if one exists)  We do not know the order, however Second chance Need reference bit Clock replacement If page to be replaced (in clock order) has reference bit = 1 then:  set reference bit 0  leave page in memory  replace next page (in clock order), subject to same rules

9.36 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Second-Chance (clock) Page-Replacement Algorithm

9.37 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Enhanced Second-Chance Algorithm Considering the reference bit and the modification bit, we have the following four classes (0,0): neither recently used nor modified (0,1): not recently used but modified (1,0): recently used but clean (1,1): recently used and modified

9.38 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Counting Algorithms Keep a counter of the number of references that have been made to each page LFU Algorithm: replaces page with smallest count MFU Algorithm: based on the argument that the page with the smallest count was probably just brought in and has yet to be used Page-Buffering Algorithms Keep a pool of free frames When page faults, the desired page is read into a free frame from the pool before the victim is written to disk Allow the process to restart as soon as possible Skip 9.4.8

9.39 Silberschatz, Galvin and Gagne ©2005 Operating System Principles The simple case: the single user system Demand paging: Besides frames for the operating system, put all other frames in the free-frame list. Variations on the simple strategy Require the OS allocate all its buffers and table space from the free-frame list. When this space is not in use by the OS, it can be used to support user paging. Keep three free frames reserved on the free-frame list at all times. 9.5 Allocation of Frames

9.40 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Allocation of Frames We cannot allocate more than the total available frames. Each process needs minimum number of pages Example: machine with all memory reference: at least two memory accesses per instruction. If indirect addressing, then paging requires at least 3 frames per process Example: IBM 370 – 6 pages to handle MVC (multiple move) instruction: instruction is 6 bytes, might span 2 pages source block might straddle 2 pages destination block might straddle 2 pages Worst case: multiple level indirection (must have a limit) Two major allocation schemes equal allocation priority allocation

9.41 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Equal Allocation Equal allocation – For example, if there are 100 frames and 5 processes, give each process 20 frames. Proportional allocation – Allocate according to the size of process

9.42 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Priority Allocation Use a proportional allocation scheme using priorities rather than size If process P i generates a page fault, select for replacement one of its frames select for replacement a frame from a process with lower priority number

9.43 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Global vs. Local Allocation Global replacement – process selects a replacement frame from the set of all frames; one process can take a frame from another Allows a high-priority process to increase its frame allocation at the expense of a low-priority process Low-priority process cannot control its own page-fault rate Local replacement – each process selects from only its own set of allocated frames Global replacement generally results in greater system throughput Skip 9.5.4

9.44 Silberschatz, Galvin and Gagne ©2005 Operating System Principles 9.6 Thrashing ( 痛擊 ) If a process does not have “enough” pages, the page- fault rate is very high. This leads to: low CPU utilization operating system thinks that it needs to increase the degree of multiprogramming another process added to the system Thrashing: a process is busy swapping pages in and out. It spends more time in paging than executing.

9.45 Silberschatz, Galvin and Gagne ©2005 Operating System Principles

9.46 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Demand Paging and Thrashing Why does demand paging work? Answer: Locality model Process migrates from one locality to another Localities may overlap A process page faults when it changes locality If we allocate fewer frames than size of the current locality, the process will thrash For a system, why does thrashing occur? sum of size of locality > total memory size We can limit the effect of thrashing by using a local replacement algorithm If processes are thrashing, the average service time for a page fault will increase because of the longer queue for the paging device

9.47 Silberschatz, Galvin and Gagne ©2005 Operating System Principles locality in a memory- reference pattern

9.48 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Working-Set Model   working-set window  a fixed number of page references Example: 10,000 instruction

9.49 Silberschatz, Galvin and Gagne ©2005 Operating System Principles WSS i (working set size of Process P i ) = total number of pages referenced in the most recent  (varies in time) if  too small will not encompass entire locality if  too large will encompass several localities if  =   will encompass entire program D =  WSS i  total demand frames of all processes if D > m (total number of available frames)  Thrashing Policy: if D > m, then suspend one of the processes

9.50 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Keeping Track of the Working Set Approximate with a fixed-interval timer interrupt + a reference bit Example:  = 10,000 and Timer interrupts after every 5000 time units Keep in memory 2 bits for each page Whenever a timer interrupts copy and sets all values of current reference bit to 0 If a page fault occurs, we can examine the current reference bit and two in-memory reference bits to determine whether a page was used during the last to references If one of the bits in memory = 1  page in working set Why is this not completely accurate? Improvement = 10 bits and interrupt every 1000 time units Why?

9.51 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Page-Fault Frequency Scheme Establish “acceptable” page-fault frequency rate If actual rate too low, process loses frame If actual rate too high, process gains frame Skip 9.7

9.52 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Working Sets and Page Fault Rates A peak in the page fault rate occurs when we begin demand-paging a new locality.

9.53 Silberschatz, Galvin and Gagne ©2005 Operating System Principles 9.7 Memory-Mapped Files (skip) Memory-mapped file I/O allows file I/O to be treated as routine memory access by mapping a disk block to a page in memory A file is initially read using demand paging. A page-sized portion of the file is read from the file system into a physical page. Subsequent reads/writes to/from the file are treated as ordinary memory accesses. Simplifies file access by treating file I/O through memory rather than read() write() system calls Also allows several processes to map the same file allowing the pages in memory to be shared Skip 9.7

9.54 Silberschatz, Galvin and Gagne ©2005 Operating System Principles

9.55 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Memory-Mapped Shared Memory in Windows Skip 9.7.2

9.56 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Memory-Mapped I/O Each I/O controller includes registers to hold commands and the data being transferred Special instructions allow data transfers between these registers and system memory Memory-mapped I/O provides more convenient access to I/O devices Ranges of memory addresses are mapped to the device registers Appropriate for devices that have fast responses, like video controllers and screens Also convenient for serial and parallel ports to connect modems and printers to a computer  By setting or clearing a bit in the device control register  Polling (programmed I/O) or interrupt driven

9.57 Silberschatz, Galvin and Gagne ©2005 Operating System Principles 9.8 Allocating Kernel Memory Treated differently from user mode memory 1. Kernel requests memory for structures of varying sizes  Some of them are less than a page in size 2. Some kernel memory needs to be contiguous.  Ex: Hardware devices interact directly with physical memory Often allocated from a free-memory pool different from the list used to satisfy ordinary user-mode processes

9.58 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Buddy System Allocates memory from fixed-size segment consisting of physically-contiguous pages Memory allocated using power-of-2 allocator Satisfies requests in units sized as power of 2 Request rounded up to next highest power of 2 When smaller allocation needed than current available, current chunk split into two buddies of next-lower power of 2  Continue until appropriate sized chunk available Advantage: adjacent buddies can be combined quickly to form larger segment using coalescing ( 聯合 ) Drawback: very likely to cause fragmentation within allocated segments

9.59 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Buddy System Allocator

9.60 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Alternate strategy: Slab Allocation Slab is one or more physically contiguous pages Cache consists of one or more slabs Single cache for each unique kernel data structure Each cache filled with objects – instantiations of the data structure When cache created, filled with objects marked as free When structures stored, objects marked as used If slab is full of used objects, next object allocated from empty slab If no empty slabs, new slab allocated Benefits include no fragmentation, fast memory request satisfaction

9.61 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Slab Allocation

9.62 Silberschatz, Galvin and Gagne ©2005 Operating System Principles 9.9 Other Issues Prepaging To reduce the large number of page faults that occurs at process startup In a system with working-set model, we keep with each process a list of pages in its working set. Before suspending a process, its working set is saved. Prepage all or some of the pages a process will need, before they are referenced But if prepaged pages are unused, I/O and memory was wasted Assume s pages are prepaged and α of the pages is used  Is cost of the s * α saved pages faults greater or less than the cost of prepaging s * (1- α) unnecessary pages?  α near zero  prepaging loses  α near one  prepaging wins

9.63 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Other Issues – Page Size Page size selection must take the following into consideration: Size of the page table  A large page size is preferred Internal fragmentation  Need a small page size Time to read or write a page  Need a larger page size Locality  Smaller page size to match program locality more accurately Number of page faults  Need a larger page size to reduce number of page faults

9.64 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Other Issues – TLB Reach TLB: expensive and power hungry TLB Reach -The amount of memory accessible from TLB TLB Reach = (TLB Size) X (Page Size) Ideally, the working set of each process is stored in TLB Otherwise there is a high degree of page faults Increase the Page Size to increase TLB reach This may lead to an increase in fragmentation as not all applications require a large page size Provide Multiple Page Sizes This allows applications that require larger page sizes the opportunity to use them without an increase in fragmentation Recent trends is to move toward software-managed TLB Other Issues – Other Issues – Inverted Page Tables An external page table must be kept for demand paging

9.65 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Other Issues – Program Structure Program structure l int[128,128] data; Each row is stored in one page Program 1 for (j = 0; j <128; j++) for (i = 0; i < 128; i++) data[i,j] = 0;  128 x 128 = 16,384 page faults Program 2 for (i = 0; i < 128; i++) for (j = 0; j < 128; j++) data[i,j] = 0;  128 page faults

9.66 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Other Issues – Program Structure Stack has good locality, hash table has bad locality In addition to locality, other factors:  search speed, total number of memory references, total number of pages touched Compiler and loader Code pages are always read-only Loader can avoid placing routines across page boundaries Routines that call each other can be packed into one page Language The use of pointers in C and C++ tend to randomize access to memory, thereby potentially diminishing a process’s locality OO programs tend to have a poor locality of reference

9.67 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Other Issues – I/O interlock (skip) I/O Interlock – Pages must sometimes be locked into memory. Consider I/O - Pages that are used for copying a file from a device must be locked from being selected for eviction by a page replacement algorithm Solutions Never execute I/O to user memory. Use system memory instead. Extra copying between user memory and system memory. Allow pages to be locked into memory with a lock bit. Lock-bit can be used in preventing replacement of a newly brought-in page until it can be used at least once. Useful for low-priority process,. Skip 9.9.6

9.68 Silberschatz, Galvin and Gagne ©2005 Operating System Principles Reason Why Frames Used For I/O Must Be In Memory Skip 9.10